AI for HR Weekly Podcast, brought to you by Barry Phillips · 2026-05-21 · 4 min
Key moments - from our scoring
Substance score
47 / 100
Five dimensions, 20 points each
The episode opens with a telling experiment: uploading an expense receipt to ChatGPT and asking it to fabricate a 20% higher amount - a request it fulfilled without hesitation. Two months later, the landscape has shifted. Phillips highlights a California courtroom ruling where a federal judge sanctioned a supervising partner for failing to verify AI-generated legal briefs containing fabricated case citations, establishing that "I didn't check it" is now the offence, not a defence. The second case involves Amazon employees gaming their internal GenAI tool called "MeshClaw" through "tokenmaxxing" - deliberately creating pointless work to inflate AI-usage metrics on internal dashboards, not to increase productivity but to manipulate performance measurements. Phillips draws a distinction between reckless misuse (trusting AI without verification) and cynical misuse (deliberately deceiving employers), arguing both require immediate policy action. He contends that HR teams must classify these behaviours clearly: failure to verify AI output as a performance issue, and deliberate AI misuse for fabrication or metric inflation as outright misconduct. The episode concludes with a hopeful note - modern AI systems are increasingly refusing unethical requests - but questions whether organizational ethics are keeping pace with machine ethics.
The supervising partner who failed to check the junior lawyer's AI-assisted brief was personally sanctioned by federal court, establishing that senior lawyers are now accountable for verifying subordinates' AI work.
Tokenmaxxing is deliberately creating pointless busywork using GenAI tools to inflate AI-usage metrics on internal dashboards, not to improve productivity but to game performance measurement systems.
Modern AI models now refuse such requests - when Phillips retried the same expense forgery prompt weeks later, the model politely but firmly declined to help.
Courts are already holding organizations legally accountable for unverified AI work, and current grey-area policies leave companies exposed while honest employees lack clear guidance on what constitutes misconduct.
Our reviewer’s read on each dimension, with quotes from the episode.
The episode delivers several concrete, actionable insights: the accountability shift in the lawyer sanctioning case (senior partners liable for subordinates' unchecked AI output), the 'tokenmaxxing' phenomenon at Amazon (gaming metrics through manufactured work), and the distinction between reckless misuse (not checking) and misconduct (deliberate deception). However, it relies heavily on anecdotes rather than systemic analysis, and the policy prescriptions (update handbooks this quarter) lack depth on implementation, audit mechanisms, or threshold-setting for different misuse categories.
senior lawyers are accountable when their subordinates misuse AI. "I didn't write it" is no longer a defence
Gaming the system by manufacturing work that no human actually needed doing, so the dashboard glows green
The framing of AI misuse as two distinct species (reckless vs. cynical) is sharper than typical 'AI risk' discourse, and the Amazon tokenmaxxing anecdote appears relatively novel to mainstream HR podcasting. However, the core argument - that organisations need clear policy on AI verification and misconduct - is fairly standard risk-management thinking. The lawyer citation case has circulated widely. The originality gains are in synthesis and specificity of examples rather than fundamental reframing.
Two cases, two species of misuse. One is reckless: trusting the machine and not checking. The other is cynical, using the machine to game your own employer
actively misusing AI in the workplace to fabricate, to inflate metrics, to deceive is a misconduct issue. Not a grey area. Not a learning opportunity. Misconduct
This is a solo monologue by Barry Phillips with no named guests. The host appears to be a commentator rather than an operator with hands-on HR experience at scale (no credentials, company affiliations, or evidence of having built or scaled HR function provided). The episode is editorial rather than interview-based.
Hello Humans! And welcome to the weekly podcast that aims to review an important AI issues in five minutes or less
The episode grounds itself in three specific, named cases: the ChatGPT expense receipt experiment (quantified: 20% bump), the California junior lawyer filing fabricated citations (federal court sanction of supervising partner), and Amazon's internal 'MeshClaw' tool with employees 'tokenmaxxing' metrics. These are concrete enough to be verifiable and memorable. However, the Amazon case lacks detail on scale, audit findings, or consequences; the lawyer case is already public knowledge; and broader data on workplace AI misuse frequency or patterns is absent.
I uploaded an expense receipt to ChatGPT and asked it to duplicate it but to bump the amount up by twenty percent
A junior lawyer filed an AI-assisted brief that contained a fabricated citation - a case that simply didn't exist
This is a solo monologue with no conversation, follow-up questions, or intellectual sparring. There is no pushback, nuance exploration, or challenge to claims. The structure is rhetorical and sermon-like rather than dialogical. For a podcast format marketed as peer-to-peer learning, the absence of genuine exchange is a material weakness.
Hello Humans! And welcome to the weekly podcast that aims to review an important AI issues in five minutes or less
Until next week, bye for now
Computed from the transcript - who did the talking, and the words that came up most.
Transcribed and scored by The B2B Podcast Index.
Hello Humans! And welcome to the weekly podcast that aims to review an important AI issues in five minutes or less. Two years ago, I tried a small experiment. I uploaded an expense receipt to ChatGPT and asked it to duplicate it but to bump the amount up by twenty percent.
No raised eyebrow, no hesitation. ChatGPT obliged in seconds. A perfect little forgery, served up with a smile. At the time, that felt like the warning shot.
The technology that could draft your emails could also help you fiddle your expenses. Earlier this month, two stories caught my eye. The first came from a California courtroom. A junior lawyer filed an AI-assisted brief that contained a fabricated citation - a case that simply didn't exist.
We've heard that one before. What was new was the judge's response. The supervising partner, who hadn't bothered to check the work, was personally sanctioned. The federal court was unambiguous: senior lawyers are accountable when their subordinates misuse AI.
"I didn't write it" is no longer a defence. "I didn't check it" is the offence. The second came from Amazon. Reports emerged that staff were using an internal GenAI tool - nicknamed "MeshClaw" - to automate pointless busywork.
Not to save time. Not to be more productive. But to inflate their AI-usage metrics. They've even coined a term for it: "tokenmaxxing."
Gaming the system by manufacturing work that no human actually needed doing, so the dashboard glows green. Two cases, two species of misuse. One is reckless: trusting the machine and not checking. The other is cynical, using the machine to game your own employer.
And both point in the same direction. The policy implications are clear, and they need to land in employee handbooks this quarter, not next year. Failing to check AI output is a performance issue. Plain and simple, if you sign your name to it, you own every word.
And actively misusing AI in the workplace to fabricate, to inflate metrics, to deceive is a misconduct issue. Not a grey area. Not a learning opportunity. Misconduct.
HR teams who haven't spelled this out yet are leaving their organisations exposed, and leaving their honest employees without the clarity they deserve. Which brings me back to my receipt. Last weekend, out of curiosity, I tried the same trick. Same prompt, same intent.
This time the model refused, point blank. Politely, but firmly, it declined to help me commit a small act of fraud. The ethics of the machines are improving. Can we say the same about ourselves?
Until next week, bye for now.
Other episodes covering the same guests and topics, from across The B2B Podcast Index.
Is Your Business Invisible to AI Search? (And How to Fix It) ft. Ray YoungRevenue Science · on ChatGPT85 / 100
Episode 018: Season 2, the $75 Consult and the Frankenstein StackAI Tools for Practicing Lawyers · on ChatGPT84 / 100
Welcome to the Software Renaissance. Your Strategy Isn't Ready | Martin ErikssonProductized Podcast · on ChatGPT82 / 100
Miles Rowland: Why Every Portfolio Company Needs an AI Engineering TeamAI Pathfinder for Private Equity Podcast · on ChatGPT81 / 100
Elevate 50: What's next for L&D? Donald TaylorLearning Uncut · on ChatGPT81 / 100
PwC's Chief People Officer on Training 80,000 People for the AI Era With Human Skills at the CenterFuture Ready Leadership With Jacob Morgan · on ChatGPT81 / 100Hello, Can I Speak to a Human, Please?51 / 100The Top Five GenAI Tools for HR57 / 100Data Centres, Water, and the Danger of Big Scary Numbers in the Workplace59 / 100The Forthcoming AI Token Crisis - What’s HR got to do with it?40 / 100Is Your Organisation AI Aware or AI Feral?37 / 100